Directed melanocyte migration: the role of Stem cell factor, cytoskeleton and focal adhesions

(1)

Thesis

Reference

Directed melanocyte migration: the role of Stem cell factor, cytoskeleton and focal adhesions

WEHRLE-HALLER, Bernhard

Abstract

Cell adhesion and migration are critical for the development of multicellular organisms and the maintenance of tissue integrity. This thesis explores the roles of the growth factor "Stem Cell factor and its receptor c-kit, in the process of cell migration and developmental patterning. In addition changes of the cytoskeleton and adhesion receptors of the family of integrins were analysed by live cell imaging and quantitative dynamic analysis. The thesis focusses on the specific contributions of the author to this field of cell adhesion, migration and morphogenesis.

WEHRLE-HALLER, Bernhard. Directed melanocyte migration: the role of Stem cell factor, cytoskeleton and focal adhesions . Thèse de privat-docent : Univ. Genève, 2002

DOI : 10.13097/archive-ouverte/unige:36681

Available at:

http://archive-ouverte.unige.ch/unige:36681

Disclaimer: layout of this document may differ from the published version.

1 / 1

(2)

Université de Genève, Faculté de Médecine

Directed melanocyte migration:

The role of Stem cell factor, cytoskeleton and focal adhesions

Bernhard Wehrle-Haller

Département de Pathologie Centre Médical Universitaire

Faculté de Médecine Université de Genève

1211 Genève 4 Suisse

Thèse d'habilitation au titre de Privat-Docent

à la Faculté de Médecine de Genève: 2002

(3)

Bernhard Wehrle-Haller: Mechanisms of Melanocyte Migration

2 Pour Monique, Noëlle et Cédric

(4)

3 Content : Page: 3

1. Introduction

1.1. Cell migration in health and desease..………..………..………..………... 4 1.2. Mechanisms of cell migration……….……….……… 6 1.3. Choosing a model system: the neural crest derived melanocyte……….………... 7 2. Directed Melanocyte Migration Induced by Growth Factor Signals

2.1. Stem Cell factor and its receptor c-kit……….…..………….. 9 2.2. Distinct activities of soluble and membrane bound Stem Cell factor………..……....….. 12 2.3. Limiting Stem Cell factor supply affects melanocyte migration and survival………….. 25 2.4. Stem Cell factor dependent migration and/or survival ?...….……….... 39 2.5. Stem Cell factor induces chemotactic migration of melanocytes……….……….… 49 2.6. Intracellular targeting of Stem Cell factor: essential roles of the cytoplasmic tail……... 63 2.7. A mono-leucine, acidic cluster associated basolateral targeting domain……..………… 80 2.8. Cytoplasmic domain of SCF: a new therapeutic target?………... 89 3. Melanocyte Migration in vitro: The Role of the Cytoskeleton

3.1. Growth factor mediated changes in the actin cytoskeleton………...…... 91 3.2. Actin dynamics during cell migration……….….... 93 3.3. Microtubule dependent rear retraction is required for directed cell migration………….. 104 4. Melanocyte Migration in vitro: Cell-Substrate Interactions

4.1. Cell migration requires dynamic remodeling of cell-substrate interactions….…………. 119 4.2. The role of the α v β 3-integrin in cell migration……….….... 121 4.3. Marching at the front dragging behind: differential α v β 3-integrin

dynamics during cell migration………...……. 122 5. Conclusions

5.1. Cross-talk between integrin, actin and microtubule cytoskeleton regulate cell adhesion 137 5.2. The inner lives of focal adhesions………..……….……… 150

5.3. Outlook………..…... 159

6. Supplementary Material and Acknowledgement

6.1. Video sequences ………..……….….. 160

6.2. Acknowledgement……..……….……….………… 161

7. References………... 162

(5)

4 1. Introduction

1.1. Cell Migration in Health and Disease

The evolutionary step from unicellular to multicellular organisms, not only required the development of adhesive systems that allowed the formation of cell layers due to homotypic cell-cell interactions, but also the ability of cells to segregate from each other and to reversibly interact with their extracellular environment. This change in cell morphology and behavior has been named epithelial to mesenchymal transformation (EMT) and is key to the development of a third germinal layer (the mesenchyme) during the process called gastrulation (Fig. 1) (Duband et al., 1995; Hay, 1995). EMT is not only crucial for gastrulation, it is implicated in the development of all organs that form by inductive mechanisms between two epithelial or an epithelial and a mesenchymal tissue.

While the changes in gene expression associated with these inductive events are in the process of being understood, the mechanisms that lead to changes of cell shape or remodeling of cell-cell or cell-substrate interactions remain elusive. Although EMT is the process that gives rise to migrating cells, continuous and efficient control of the segregated cells is paramount to the survival of the entire organism.

The immune system is another example where cell migration is of importance. Millions of cells that constantly patrol to fight possible intruder, depend on efficient adhesion mechanism that are regulated by soluble or cell surface bound signaling proteins of the chemokine family (Baggiolini, 2001). In addition, after tissue damage, cleaning of cellular debris and recruitment of new cells by proliferation and migration requires very efficient control over the motile apparatus of each cell (Fig.

1).

Since cell motility is paramount to immune surveillance, tissue regeneration and homeostasis, pathological conditions are frequently associated with the loss of regulated cell migration. While in some situations, like during the formation of metastasis, the molecular causes that lead to the dispersion of tumor cells are not well understood, there has been progress in the identification of genes, that if mutated, can cause the reduction of cell adhesion and migration of cells of the immune system. Such conditions are often reflected by the failure of the immune or the hematopoietic system to function properly and have resulted in significant progress in the understanding of cell migration.

For example, the Wiskott Aldrich Symptome protein (WASP) has been identified as an important

regulator of the actin cytoskeleton of migrating macrophages, while defects in platelet glycoprotein

GPIIb/IIIa, a member of the integrin family of cell-substrate receptors, results in bleeding disorders

(6)

5 such as the Glanzmann disease (Badolato et al., 1998; Chen et al., 1992; Jones, 2000; Nurden and Nurden, 2001). In other cases, naturally occurring mutations of growth factors or their receptors, and the specific knockout of signaling molecules of the chemokine receptor family have demonstrated the strict requirement of specific subsets of migratory cells on these signals for survival and migration (Debard et al., 1999; Dietrich et al., 1999; Forster et al., 1996; Schuchardt et al., 1994).

Here in this thesis, I will concentrate on the neural crest derived melanocyte lineage and show, how genetic analysis has helped to identify key proteins and cellular components required for melanocyte migration. In addition, these genetic studies are complemented by an in vitro approach to quantitatively analyze the role of crucial structural proteins in directed melanocyte migration.

I will demonstrate here, that despite limited information about the mechanisms of cell migration, the development of new therapeutic strategies targeting the various cellular and extracellular components involved in the regulation of motility may have an impact on the treatment of melanoma, one of the most aggressive type of cancers, and other pathological situations involving insufficient or superfluous cell motility.

Fig. 1. Cell migration is implicated and plays a major role during development (e.g. gastrulation), tissue

homeostasis, wound-healing, immune surveillance (represented by diapedesis) and pathological conditions such

as metastasis formation.

(7)

6 1.2. Mechanisms of Cell Migration

Although cell migration is an extremely complex biological phenomenon, it requires three basic ingredients (Fig 2). First, attractive or repulsive signals stimulate migration and guide cells through the tissue. Growth factors, chemokines, or bacterial products mediate attractive or chemotactic signals. In contrast, repulsive behavior is induced by ligands of the Ephrin family (Krull et al., 1997; Wang and Anderson, 1997; Wilkinson, 2000). The second component is the cell, which captures these signals with specific cell surface receptors, that often belong to the family of receptor tyrosine kinase (RTK) or G-protein coupled receptors. Whether a signal is attractive or repulsive depends on the how the cell is able to transform this information into cell shape changes. Specifically, forward movement is generated by the polymerization of the actin cytoskeleton at specific sites in the cell called lamellipodia or filopodia (Ridley et al., 1992; Small et al., 2002). In contrast, repulsive signals induce the actin-myosin dependent contraction of the polymerized actin cytoskeleton leading to the retraction of cellular processes (Ridley and Hall, 1992; Wahl et al., 2000). However, without the third component, represented by the physical interaction with the extracellular environment, a cell will not be able to transform cytoskeletal changes into motility. In order to reversibly link the polymerized actin cytoskeleton to the extracellular matrix, the cell employs heterodimeric transmembrane receptors of the integrin family (Hynes, 1992). The coordinated interactions of these three components determine when, how and where a cell will migrate.

Here I will first analyze the individual role of these three different components during melanocyte migration and will subsequently link them to a comprehensive model of cell migration.

Fig. 2. The process of cell migration can be reduced to three essential components. A signal (1) stimulates and

gives directional information for migration. The cell (2) provides receptors for these signals, which result in the

specific reorganization of the cytoskeleton. The extracellular substrate (3) provides physical anchor points, in

order to transform cell shape changes into motility.

(8)

7 1.3. Choosing a Model System: Neural Crest Derived Melanocytes

As mentioned above, migrating cells can be isolated from almost all organs and tissues.

However, dependent on their origin and initial function, the manner in which cells migrate can vary extensively. For example, polymorphonuclear neutrophils (PMN) derived from the immune system migrate very fast towards a source of chemokine, maintaining a globular shape with a short protruding trail also called uropode (Niggli, 1999). While these cells hardly touch the substrate during migration, equally fast migrating fish keratinocytes exhibit large fan shaped flat lamella, which give the impression of an elegant gliding over the substrate (Lee and Jacobson, 1997; Lee et al., 1994). Due to their regular and flat lamellipodia these cells are ideally suited for histochemical analysis (Verkhovsky et al., 1999). On the other hand, slow moving fibroblasts can be genetically manipulated with ease, however advance by cycles of lamellipodia extension and retraction (Lauffenburger and Horwitz, 1996).

The ideal model system for the study of cell migration should combine all of the above features such as (i) directed migration in response to specific signaling molecules, (ii) flat and fast advancing lamellipodia that allow precise microscopic analysis and (iii) the accessibility to genetic and biochemical manipulations. Neural crest derived cells of the melanocyte lineage fulfill these requirements and are therefore ideally suited to analyze cell migration at the tissue as well as the single cell level.

The neural crest is specified by and forms between the neural plate and the ectoderm. During

neural tube closure, cells from the dorsal neural tube are undergoing EMT and move into the

extracellular matrix filled space bordered by the somites, ectoderm and neural tube, called the

migration staging area (MSA) (Derby, 1978; Duband et al., 1995; Erickson et al., 1992; Loring and

Erickson, 1987; Weston, 1991). The emigrated cells behave as stem cells migrating along stereotypic

pathways to join their specific destinations where they are giving rise to peripheral neurons and glia,

connective tissue and melanocytes (Le Douarin and Kalcheim, 1999; Selleck et al., 1993; Stemple

and Anderson, 1993; Wehrle-Haller and Weston, 1997) (Fig. 3). Specifically, cells that will give rise

to neural crest derived melanocytes migrate along the dorsolateral pathway that is bordered by the

dermamyotome and the overlaying ectoderm. After a first phase of migration through the

mesenchymal dermis, melanocyte precursors penetrate the basement membrane of the overlaying

epidermis and continue to disperse over the entire surface of the embryo on the epidermal side of this

basement membrane (Mackenzie et al., 1997; Nishikawa et al., 1991; Wehrle-Haller and Weston,

1995). After embryonic day 15 of mouse development, melanocyte precursors are beginning to

migrate towards and into the forming hair bulbs (Jordan and Jackson, 2000b). Hair bulbs that fail to

(9)

8 get populated by the spreading melanocyte precursors will remain un-pigmented throughout live.

Due to the precise timing and pattern of this migration, mouse mutations that affect coat pigmentation can give valuable information about the mechanism and the regulation of melanocyte migration (Jordan and Beermann, 2000). In chapter 2, I will discuss how the analysis of the Steel (Mgf

^Sl

) and Dominant White Spotting (Kit

^W

) mouse mutations that code for Stem Cell factor (also known as MGF, mast cell growth factor; KL, kit-ligand) and its respective receptor tyrosine kinase (RTK) c-kit, allowed to understand the regulation of melanocyte precursor migration by external signals in great detail.

In addition to the analysis of melanocyte migration in vivo, melanocyte precursors or transformed melanoma cells can be easily cultured and analyzed by live microscopy in vitro. While melanoma cells constitutively migrate by employing large flat lamellipodia, c-kit expressing melanocyte precursors (Sviderskaya et al., 1995) can be stimulated to migrate in response to added Stem Cell factor (Ballestrem et al., 2000). Since these cells can be transfected with wildtype or mutant forms of green fluorescent protein tagged proteins, the role of particular proteins in cell migration can be analyzed at a cellular or sub-cellular level. In chapter 3 and 4, experimental data is presented, that demonstrate the role of the actin and microtubule cytoskeleton in melanocyte migration and how migration dependent changes of the actin cytoskeleton are transmitted to the underlying extracellular matrix through heterodimeric receptors of the integrin family (Hynes, 1992).

Fig. 3. Graphic representation of the different

stages of neural crest cell migration in the trunk.

Neural crest cells (NC) emigrate from the dorsal neural tube (NT) and transiently accumulate in the migration staging area (MSA). The

"intersomitic pathway" (black arrow, upper graphic) is used prior to the segregation of the spherical somites into a medial sclerotome and dorsolateral dermamyotome (D/M). Neural crest cells subsequently invade the rostral portion of the sclerotome (RS) on the "medial pathway"

(black arrow, middle graphic), reaching locations where sympathic and sensory ganglia will form.

Neural crest cells will migrate along the dorso- lateral pathway (black arrow, lower graphic) that forms between the dermamyotome and the overlying ectoderm (ECT), almost a day later compared to the medial pathway. Neural crest cells migrating on this pathway will mainly develop into melanocytes, subsequently dispersing all over the surface of the embryo.

Notochord (n), caudal somite (CS).

(10)

9 2. Directed Melanocyte Migration Induced by Growth Factor Signals 2.1. Stem Cell factor and its receptor c-kit

Pioneering work by Elisabeth Russell had led to the discovery of mouse mutants that specifically affected the fate and survival of neural crest derived melanocytes, but not melanocytes derived from the central nervous system (CNS), resulting in the characteristic black-eyed/white-coat phenotype of homozygous Steel and Dominant White Spotting mutants (Russell, 1979). This is in contrast to mutations of the tyrosinase gene which block the synthesis of melanin in neural crest and CNS derived melanocytes resulting in the albino phenotype (Silvers, 1979). In addition to the absence of melanocytes in the skin and hair, Steel and Dominant White Spotting mutant mice display also defects in their germ cells resulting in sterility, a reduction of hematopoietic stem cells in the bone marrow leading to anemia and a lack of peripheral mastocytes in the tissues (Morrison-Graham and Takahashi, 1993; Williams et al., 1992). While the genetic defect in the Steel mutation was traced to the tissue environment in which the affected cell populations were localized, the Dominant White Spotting mutation caused the loss of these cell populations even when transplanted into a normal environment (Russell, 1979). In 1990 several laboratories simultaneously identified the gene affected by the Steel mutation, encoding for a membrane-bound growth factor named Stem Cell factor (SCF) but also called (Mast cell growth factor, MGF or kit-ligand, KL) (Anderson et al., 1990; Copeland et al., 1990; Flanagan and Leder, 1990; Huang et al., 1990; Martin et al., 1990; Williams et al., 1990;

Zsebo et al., 1990a; Zsebo et al., 1990b). In contrast, the receptor tyrosine kinase (RTK) c-Kit is

encoded at the Dominant White Spotting locus and is structurally related to the RTK's c-fms, PDGF

receptor alpha and beta, arranged pair wise on two related chromosomes (PDGFR α /c-kit and

PDGFR β /c-fms) (Kataoka et al., 1997; Qiu et al., 1988; Yarden et al., 1986; Yarden et al., 1987). The

fact that c-Kit originates from a duplication of the PDGFR locus during the evolution of the

vertebrates sheds some light on the evolutionary mechanisms involved, in order to create new cell

populations with novel functions to increase the complexity and capacities of new organisms (see

below). SCF is most homologous to the macrophage colony-stimulating factor (M-CSF or CSF-1),

which binds to its receptor c-fms (Bazan, 1991). Similar to CSF-1, SCF is expressed as a dimeric

membrane-bound precursor generated from two differentially spliced transcripts of which the longer

one is readily cleaved by cell surface proteases to generate soluble SCF (Huang et al., 1992). The

complex of SCF bound to c-Kit has recently been crystallized and reveals the common structure of a

bundle of four α -helices linked by two intra-molecular disulfide bridges (Fig. 4) (Bazan, 1991; Jiang

et al., 2000; Zhang et al., 2000). An interest in the differentially spliced forms of SCF and its

(11)

10 transmembrane and cytoplasmic domains has been generated due to the molecular characterization of a naturally occurring Steel allele (Steel-dickie) (Mgf

^Sl-d

) exhibiting a phenotype almost as severe as the null mutations (Brannan et al., 1991; Flanagan et al., 1991). The Mgf

^Sl-d

allele represents a deletion of the transmembrane and cytoplasmic domain of SCF, resulting in the secretion of a fully functional form of SCF from cells transfected with the mutant SCF construct. Based on these data it was speculated that the membrane bound splice variant of SCF that lacks the major proteolytic cleavage site (Huang et al., 1992) is responsible for the survival of c-Kit expressing SCF-dependent cell populations in vivo (Brannan et al., 1991; Flanagan et al., 1991). Consequently, we (J.A. Weston and myself) began to speculate whether the soluble or proteolytically shed form of SCF may have a chemotactic function in attracting or guiding c-kit expressing cell populations such as melanocytes, that have been recognized to extensively migrate through the embryonic and adult tissue.

Particularly, it has been demonstrated in vitro, that PDGFR α , β , c-Kit and c-fms expressing motile cells show chemotactic responses towards concentration gradients of their ligands (Allen et al., 1997; Blume-Jensen et al., 1991; Kundra et al., 1995; Ueda et al., 2002). In addition, a c-Kit expressing small cell lung carcinoma cell line responds chemotactically to SCF in vitro (Sekido et al., 1993). Mast cells, which express high levels of c-Kit on their surface, respond to SCF stimulation by spreading on adhesive substrates. However, a 100-fold higher SCF concentration is required to obtain proliferation of these c-Kit expressing mast cells (Dastych and Metcalfe, 1994; Kinashi and Springer, 1994). These results suggest that both motogenic and mitogenic responses can be induced by SCF presentation towards responsive cells.

We conducted several experimental approaches to understand the role of SCF in melanocyte precursor migration in order to elucidate the role of the soluble and membrane bound forms of SCF (chapter 2.2). In addition, the presence of SCF proofed to be critical for melanocyte precursor migration and survival on the lateral pathway (chapter 2.3), which led us to hypothesize that growth factor dependent survival and chemotactic activities direct the migration and specific localization of other neural crest cell derivatives (chapter 2.4). In chapter 2.5 we provided the proof that SCF is the only chemotactic cue generated on the lateral neural crest migration pathway that is able to induce migration of melanocyte precursors in vivo. These data are complemented by studies on the role of the cytoplasmic tail of SCF for efficient presentation towards c-Kit expressing cells such as migrating melanocyte precursors (chapter 2.6 and 2.7). Furthermore, a short chapter discusses the potential role of pathological changes in SCF localization and production, and the respective changes in the behavior of c-kit expressing responsive cells of the mastocyte and melanocyte lineage (chapter 2.8).

In addition, this chapter contains a paragraph suggesting, that SCF localization and expression may

(12)

11 provide a new therapeutic target to reduce hyperpigmented lesions in the skin and/or allergic reactions.

.

Fig. 4. Crystal structure of head to head dimerized SCF. Figure adapted from Jiang et al., 2001.

(A) Ribbon diagram. (B) C

^α

stereodiagram of the AB dimer. Note that the bundle of four α -helices

joined by two beta-strands and intra-molecularly cross-linked by two S-S bridges (yellow). The linker

region and C-terminal sequences are not part of the crystal structure.

(13)

12 2.2. Distinct activities of soluble and membrane bound Stem Cell factor

(14)

INTRODUCTION

In the trunk of vertebrate embryos, neural crest cells segregate from the neural epithelium and transiently reside in a

‘migrating staging area’ (MSA) delimited by the neural tube, the somite and the overlying epithelium. Crest cells immedi- ately begin to disperse from the MSA on a ventromedial pathway along the myotome and into rostral sclerotomal mes- enchyme (Erickson and Loring, 1987; Weston, 1991). Later, crest cells remaining in the MSA disperse on a dorsolateral pathway, migrating between the dermatome and the epidermis (Erickson et al., 1992). The crest cells on the lateral pathway subsequently become interspersed with dermal mesenchyme and cross the epithelial basement membrane to localize in the epidermis. In the mouse, these cells remain in the epithelium for many days before they undergo melanogenesis postnatally.

Before early markers for pigment cell precursors were available, the timing of melanocyte precursor dispersal and localization in the skin was inferred by testing for the ability of cultured or grafted tissue to produce melanocytes (Rawles

1947; Derby 1978). Results of such studies revealed that the first melanocyte precursors are present on the lateral pathway of the embryonic trunk at about e11 and reach the limb buds by e12. At a lateral trunk level most of the melanocyte pre- cursors enter the epidermis between e13 and e14 (Mayer, 1973). Recently, histochemical reagents that intensify pigment in melanosomes of otherwise undifferentiated melanocytes, or probes for melanocyte markers such as the tyrosine kinase receptor, c-kit, and tyrosinase-related protein-2 (TRP-2) confirmed these inferences, and revealed the presence of melanocyte precursors in the head as early as e10.5 (Manova and Bachvarova, 1991; Steel et al., 1992; Pavan and Tilghman, 1994).

Several mouse mutations affecting coat pigmentation have been described and their molecular defects characterized. Two of the best studied, Steel (Sl) and Dominant spotting (W), are embryonic lethals as homozygotes, due to failure of erythro- poiesis, whereas heterozygous embryos are viable but eventu- ally show a white spotting coat color pattern. The defective gene in W mutants codes for a receptor tyrosine kinase (c-kit;

731

Development 121, 731-742 (1995)

Printed in Great Britain © The Company of Biologists Limited 1995

Trunk neural crest cells segregate from the neuroepithe- lium and enter a ‘migration staging area’ lateral to the embryonic neural tube. After some crest cells in the migration staging area have begun to migrate on a medial pathway, a subpopulation of crest-derived cells remaining in the migration staging area expresses mRNAs for the receptor tyrosine kinase, c-kit, and tyrosinase-related protein-2, both of which are characteristic of melanocyte precursors. These putative melanocyte precursors are sub- sequently observed on the lateral crest migration pathway between the dermatome and overlying epithelium, and then dispersed in nascent dermal mesenchyme.

Melanocyte precursors transiently require the c-kit ligand, Steel factor for survival. Although Steel factor mRNA is transiently expressed in the dorsal dermatome before the onset of trunk neural crest cell dispersal on the lateral pathway, it is no longer produced by dermatomal cells when melanocyte precursors have dispersed in the dermal mesenchyme. To assess the role of Steel factor in

migration of melanocyte precursors on the lateral pathway, we analyzed melanocyte precursor dispersal and fate on the lateral pathway of two different Sl mutants, Sl, a null allele, and Sl

^d

, which lacks cell surface-associated Steel factor but produces a soluble form. No melanocyte precursors were detected in the dermatome of embryos homozygous for the Sl allele or in W mutants that lack functional c-kit. In contrast, in embryos homozygous for the Sl

^d

allele, melanocyte precursors appeared on the lateral pathway, but subsequently disappear from the dermis. These results suggest that soluble Steel factor is required for melanocyte precursor dispersal on the lateral pathway, or for their initial survival in the migration staging area. In contrast, membrane-bound Steel factor appears to promote melanocyte precursor survival in the dermis.

Key words: Steel factor, melanocyte precursor, neural crest, c-kit, TRP-2

SUMMARY

Soluble and cell-bound forms of steel factor activity play distinct roles in melanocyte precursor dispersal and survival on the lateral neural crest migration pathway

Bernhard Wehrle-Haller and James A. Weston

Institute of Neuroscience, University of Oregon, Eugene, OR 97403, USA

Email: Weston@UONEURO.UOREGON.EDU

13

(15)

732 Geissler et al., 1988) and its growth factor ligand, variously called Steel Factor (SlF), Stem cell factor (SCF), mast cell growth factor (MCF) or kit ligand (KL) (Anderson et al., 1990;

Copeland et al., 1990; Huang et al., 1990; Nocka et al., 1990a;

Williams et al., 1990; Flanagan and Leder, 1990) is encoded at the Sl locus. For both genes, less severe alleles have been described that are viable as homozygotes, but show a white coat, anaemia and sterility (see Morrison-Graham and Takahashi, 1993; Copeland et al., 1990; Flanagan et al., 1991;

Brannan et al., 1991; Duttlinger et al., 1993).

A number of in vivo and in vitro approaches have been initiated to pinpoint the critical period when SlF or c-kit activity is required for survival of pigment precursors. For example, an early function for c-kit and SlF is suggested by expression of c-kit mRNA in cells found dorsal and lateral to the somites (Manova and Bachvarova, 1991) and the presence of SlF mRNA in the somitic dermatome at around e10.5 (Matsui et al., 1990). In utero injection of c-kit blocking antibody has revealed an early (e10.5), an intermediate (e14.5) and a late effect of c-kit function on melanogenesis (Nishikawa et al., 1991; Yoshida et al., 1993). Consistent with these infer- ences, analysis of melanogenesis in vitro has shown that SlF is transiently required for melanogenic precursor survival between day 2 and 6.5 in culture (equivalent to e11.5-16;

Morrison-Graham and Weston, 1993).

SlF is normally expressed in two splice-variants, both of which are initially localized to the cell surface. The larger variant contains an extracellular proteolytic cleavage site, which permits release of SlF from the cell surface. The smaller splice-variant is usually not cleaved and normally remains associated with the cell surface (Flanagan et al., 1991; Huang et al., 1992; see also Williams et al., 1992). Although frag- mentary distribution and mutational analysis of the influence of the c-kit/SlF system on melanocyte precursors exists (Steel et al., 1992), it is not yet clear whether the functions of alter- natively spliced SlF are different, or even what such functions might be.

An opportunity to address this issue is provided by assessing the early morphogenetic behavior and fate of melanocyte pre- cursors in embryos carrying various mutations at the Steel locus. For example, in the Steel-dickie (Sl

^d

) allele, the trans- membrane and cytoplasmic domains of SlF are deleted so that only a secreted form of SlF is produced (Flanagan et al., 1991;

Brannan et al., 1991). In spite of the presence of SlF activity (Brannan et al., 1991), mice heterozygous for Sl

^d

exhibit a distinct coat color phenotype. Embryos homozygous for this allele are viable but exhibit characteristic phenotypes including lack of coat pigmentation. In these mutants, melanocyte pre- cursors are initially present in the head but fail to survive (Steel et al., 1992). It is not yet known, however, how these pheno- types arise, or if the migration behavior and growth factor requirements of trunk and head melanocyte precursors are different.

In the present study, we have examined the early dispersal and fate of melanocyte precursors in Sl (null) and Sl

^d

mutants to elucidate the function of SlF in determining the early pigment patterns, and to distinguish the role(s) of the soluble and cell-bound forms of SlF in regulating the early migration behavior of pigment cell precursors. Thus, we have character- ized melanocyte precursor appearance in the MSA, and their dispersal and fate in relation to the time and location of SlF

mRNA expression in normal embryos and embryos homozy- gous for two Steel alleles. Our results indicate that migrating melanocyte precursors respond to cues provided by diffusible (soluble) SlF, through activation of the c-kit receptor. We conclude that soluble SlF is sufficient for responsive melanoblast precursors to initiate their dispersal onto the lateral pathway in vivo, but that cell-bound SlF is necessary for sub- sequent survival of pigment cells in the newly formed dermal mesenchyme.

MATERIALS AND METHODS

Genotyping mouse embryos

Inbred colonies of B6, B6Sl^d, WBReSl and WBW were maintained in our laboratory. Dated matings were set up in the evening and plugs checked the following morning. The presence of a plug was consid- ered as e0.5.

Since Sl, Sl^dand W homozygous are sterile, we mated heterozy- gous mice to obtain homozygous embryos. Sl, Sl^dor W homozygous embryos can only be recognized by morphological criteria after about e15.5, when their liver appears pale compared to wild-type or `het- erozygous littermates. In order to identify the genotype of the Sl and Sl^dembryos earlier, we amplified a genomic sequence present in the Steel gene, which is deleted in the Sl (Copeland et al., 1990) and Sl^d (Flanagan et al., 1991) mutations. According to Steel et al. (1992) we used a set of primers from the 7th and 8th exon respectively to amplify a 700 bp fragment containing mainly intron sequences. Genomic DNA was isolated from excised limb buds by proteinase K and 0.5%

SDS extraction and ethanol precipitation. Following an amplification with 35 cycles (45 seconds at 93°C, 1 minute at 56°C, 1 minute at 72°C) a 700 bp fragment was detected on a 1% agarose gel. Both Sl/Sl and Sl^d/Sl^d embryos could be identified by the absence of that amplified band. As a control for the quality of the genomic DNA as well as the amplification reaction (PCR), we used an alternative set of primers, amplifying a 500 bp genomic fragment from the gene coding for PDGFRαlocated on chromosome 5 (Orr-Urtreger et al., 1992). We performed separate amplifications with the same template DNA or included both sets of primers in one reaction. The following primers were used: Sl forward primer CCATG- GCATTGCCGGCTCTC (bases 665 to 684; Huang et al., 1990) and Sl reverse primer CTGCCCTTGTAAGACTTGACTG (complement of bases 757 to 736). PDGFRαforward primer ACCTCCTTTCG- GACGATGAC (bases 2417 to 2436; Wang et al., 1990, Acc#

M57683) located within the interkinase region and a reversed genomic primer corresponding to a sequence located within an intron 500 bp apart ATCACTTCAGAATGGCTCCA (Peter Lonai; personal communication).

Embryos homozygous for the W gene, could not be identified using a PCR approach since the W (null) allele is a point mutation in a splice site (Hayashi et al., 1991), which generates a nonfunctional c-kit protein (Nocka et al., 1990b). Therefore, all embryos from a litter produced by mating heterozygotes were treated identically, and their mutant phenotype assessed by the localization of melanocyte precursors (see below). As expected, a quarter of all the embryos scored by PCR were homozygous for either Sl or Sl^d, or were identified as W homozygotes by their pigment pattern.

cDNA probes and antibodies

cDNAs for c-kit were kindly provided by Dr Robert J. Arceci (Boston), for Steel factor (KL-M1) by Dr John Flanagan (Boston) and TRP-2 by Dr Ian Jackson (London). Polyclonal anti-fibronectin and laminin were obtained from Collaborative Research. Rhodamine goat anti-rabbit was from Cappel.

Digoxigenin (DIG)-labeled sense and antisense riboprobes were

B. Wehrle-Haller and J. A. Weston

14

(16)

733 Roles of SlF in melanocyte dispersal and survival

generated using a standard protocol (Boehringer Mannheim) from lin- earized plasmids and used for whole-mount in situ hybridization at concentrations of 1 µg/ml. RNA synthesis was checked by agarose gel electrophoresis.

Whole-mount in situ hybridization and antibody staining Whole-mount in situ hybridization was performed according to Yamaguchi et al. (1992; Y. Takahashi, personal comm.). The hybridization of DIG-labeled single-stranded RNA probes was detected with anti-DIG antibodies coupled to alkaline phosphatase. Details of the protocol will be provided upon request.

As a control for the specificity of our in situ protocol, we hybridized with sense riboprobes for KL-M1 and c-kit under identical conditions.

We were not able to generate a sense probe from our plasmid containing the TRP-2 probe. No staining could be detected in embryos hybridized with any of the sense probes. In experimental and control embryos, occasionally unspecific stain could be detected in the lumen of the spinal cord, otic vesicle and brain vesicles. Puncture of the hindbrain did reduce this unspecific trapping of stain. Embryos from e9.5 to e13.5 were treated identically with the exception of prolong- ing the proteinase K treatment for older embryos to enhance penetration of the probes. It must be emphasized that complete penetration could only be achieved with embryos of e9.5 to e10.5. In older embryos the detection of mRNAs was limited to the surface of the embryos. Fig. 3E roughly indicates the limit of penetration of the probes and antibodies. Sensitivity of the whole-mount in situ hybridization was comparable to in situs on sectioned material, since we were able to visualize every previously published source of SlF and c-kit mRNA in e9.5 to e10.5 embryos. However, compared to hybridization of tissue sections, the temporal and spatial resolution of the whole-mount in situ hybridizations is superior.

For immunohistochemistry, whole-mount in situ hybridized embryos were equilibrated in 20% sucrose and embedded in Tissue Tec, frozen and sectioned at 16 µm. Sections were dried on gelatin/alum-subbed slides and blocked with 0.5% BSA in PBT, incubated for 45 minutes with anti-fibronectin or anti-laminin rabbit antisera at 1/50 dilution in 0.5% BSA in PBT. After washing, secondary rhodamine-labeled goat anti-rabbit was added for 45 minutes diluted 1/100 in 0.5% BSA in PBT. Sections were washed and embedded in glycerol containing 1% n-propylgallate.

RESULTS

Cells expressing c-kit and TRP-2 mRNA (melanocyte precursors) in the migration staging area disperse on the dorsolateral pathway in a rostral-caudal sequence

As in previous reports (Keshet et al., 1991; Motro et al., 1991;

Orr-Urtreger et al., 1990), cells expressing c-kit mRNA were observed in the head and trunk of e9.5 embryos (not shown).

Although whole-mount in situ hybridization revealed c-kit mRNA in cells at various locations of e9.5 embryos, their neural crest origin could not be established. No hybridization signal was revealed by using the TRP-2 riboprobe (not shown), so melanocyte precursors could not be unequivocally identified.

By e10.5, cells located in sites usually occupied by migrating cranial crest cells can be seen to express the TRP-2 mRNA. These cells were detected between the forebrain- midbrain junction and the eye (Fig. 1B), at the level of the midbrain-hindbrain junction, and posterior to the otic vesicle.

Cells expressing c-kit mRNA were also detected in these locations (Fig. 1A). Based on the time of appearance and their

locations (compare with Steel et al., 1992), cells expressing both mRNAs seem likely to be the earliest crest-derived melanocyte precursors. It should be emphasized, however, that at this stage, c-kit mRNA-expressing cells were also detected in the olfactory pit, and the clefts of branchial arches I-IV, where TRP-2 mRNA expression was not detected (Fig. 1A,B).

At this stage, only a few cells expressing TRP-2 or c-kit mRNA were detected at more posterior axial levels between the otic vesicle and the most rostral somites (Fig. 1A,B).

At e11, localization of TRP-2 and c-kit mRNA-expressing cells in the head is dramatically different. These cells were present in the head mesenchyme particularly in the region between the eye, forebrain and nose, and at the posterior regions of the branchial and hyoid arch. In addition, many TRP-2 as well as c-kit mRNA-expressing cells are localized in a stripe originating at the mid-to-hindbrain junction and extending towards the eye (Fig. 1E,F; arrowheads). In contrast, no TRP-2 mRNA-expressing cells could be detected in facial structures. Within the trunk, cells expressing TRP-2 mRNA were observed from the first somites to the level of the hind limb buds (Fig. 1D) in the same locations as cells expressing c-kit mRNA (Fig. 1C). These cells were variously localized in the migration staging area (MSA) just lateral to the neural tube at posterior axial levels, or further lateral over the somites at more anterior levels (Fig. 1C,D). Groups of TRP-2 or c-kit mRNA-expressing cells were present in a segmented pattern along the axis from very rostral to mid trunk levels. In addition, c-kit mRNA-positive cells were detected in the lateral mes- enchyme, the dorsal neural tube (arrow in Fig. 1E), limb buds, and the posterior gut where no TRP-2 mRNA-expressing cells could be detected (Fig. 1D).

By e11.5, as previously reported (Steel et al., 1992), TRP- 2 mRNA-expressing cells were detected at locations in the head mesenchyme anterior and posterior to the eye, over the hindbrain, around the otic vesicle and at the posterior aspect of the hyoid arch (Fig. 2B). In the trunk, all cells expressing TRP-2 mRNA were found distributed over the entire somite surface, but no cells were detected in the lateral mesenchyme and in the limb buds (Fig. 2A). In contrast, individual c-kit mRNA-expressing cells were found dispersed over the somite surface, but also more lateral within the lateral mesenchyme (Fig. 2C). This cell population never expressed TRP-2 mRNA. We do not yet know if these cells are crest-derived or represent a mesodermally derived population, which will require more specific probes to distinguish. Expression of c- kit mRNA in mesenchymal cells was similar to that seen at earlier stages (Fig. 1C). More caudally, at the base of the tail, cells expressing TRP-2 mRNA were localized in the MSA lateral to the neural tube, and just one or two somites more rostral cells were found over the somite lateral to the MSA (Fig. 2D). In progressively older (more rostral) somites, incrementally more cells were localized laterally, and fewer TRP-2 mRNA-expressing cells remained near the neural tube (Fig. 2E). In embryos of this age, TRP-2 mRNA-expressing cells were first observed at the base of the hind limb bud (Fig.

2E).

The temporal differences between developing segments are most distinct at the level of the hindlimb of e11.5 embryos.

Accordingly, this axial region best displayed the various stages of the migration of crest cells on the lateral pathway.

Therefore, embryos that had been stained as whole mounts

15

(17)

734 for TRP-2 mRNA were sectioned trans- versely and stained with antibodies against laminin and fibronectin to reveal the archi- tecture of the somites (see Methods). Thus, Fig. 3A shows a bright-field view of a section at the base of the tail where two groups of TRP-2 mRNA-expressing cells can be seen (corresponding to A-B in Fig.

2D). One group of cells resides in the MSA lateral to the neural tube, and the other group was localized within the dermatome.

The staining for laminin revealed the borders of the neural tube and the der- matomal epithelial tissue (Fig. 3B). A section of a different embryo at a slightly older (more rostral) axial level (corre- sponding to C-D in Fig. 2D) shows elongated TRP-2 mRNA-positive cells on the laminin-positive basement membrane (arrows in Fig. 3C,D). At an even more rostral level (corresponding to E-F in Fig.

2D), the epithelial dermatome has already transformed into the mesenchymal dermis, made obvious by the intense fibronectin staining (Fig. 3F). TRP-2 mRNA contain- ing cells are evenly spread throughout the dermal mesenchyme at this axial level (see also Pavan and Tilghman, 1994, Fig. 2a).

By e12.5, cells expressing TRP-2 and c- kit mRNA were detected bordering the whisker fields and localized over the nose (Fig. 4C). The regions around the ear were also densely populated by TRP-2 mRNA- expressing cells (Fig. 4A-C). In the trunk at this stage, TRP-2 mRNA-expressing cells were present between the neural tube and the lateral midline, and some cells were seen as far ventrad as the base of the forelimb bud (Fig. 4A,B,D). No TRP-2 mRNA was detected on the ventral side of the body (Fig.

4D). Many TRP-2 mRNA-expressing cells were present on the posterior side of the hindlimb (Fig. 4C, arrow) as well as on the lateral sides of the tail.

Steel factor mRNA is expressed in a rostrocaudal sequence in the dorsal portion of the dermatome in e10.5- 12.5 mouse embryos

At e9.5, whole-mount in situ hybridization reveals SlF mRNA expression in various places in the embryo: the head mesenchyme, branchial arches, the mesenchyme posterior to the last arch, part of the gut, the tail and the kidney primordium. At this stage, no SlF mRNA could be detected within the somitic tissue of the trunk (not shown).

In the head of e10.5 embryos, SlF mRNA was detected in the mesenchyme between the telencephalon and the olfactory pit, and in the mesenchyme surrounding the otic B. Wehrle-Haller and J. A. Weston

Fig. 1. c-kit and TRP-2 mRNA-expressing cells show a similar distribution pattern in e10.5 and e11 embryos. c-kit (left panel) and TRP-2 (right panel) antisense whole- mount in situ hybridization of e10.5 (A,B) and e11 (C,D) mouse embryos. (Note almost identical localization of punctate staining at arrowheads between c-kit and TRP-2 hybridized embryos, and the absence of forelimb and hindlimb in C and D,

respectively.) The arrow in E points to c-kit mRNA expression in the spinal cord. ba, first branchial arch; e, eye; fb, forebrain; fl, forelimb; ha, hyoid arch; hb, hindbrain; hl, hindlimb; mb, midbrain; op, olfactory pit; ov, otic vesicle. Bar, 400 µm.

16

(18)

735 Roles of SlF in melanocyte dispersal and survival

vesicle (see also Steel et al., 1992). SlF mRNA was also detected in the first and second branchial arches at their most posterior proximal edges (arrowheads in Fig. 5B). In the trunk of e10.5 embryos, the dermatomes express SlF mRNA in their dorsal portions. In slightly less-advanced embryos at that gestational stage, SlF mRNA was detected in der- matomes from the first somite to the midtrunk level (Fig. 5A), whereas in developmentally more advanced embryos, SlF mRNA expression was detected in dermatomes from the first somite back to somites at the hindlimb level. Transverse sections of whole-mount embryos counterstained with a poly- clonal antiserum against laminin (Fig. 5K) or fibronectin (Fig.

5H) clearly localizes the SlF mRNA to the dorsal epithelial dermatome (Fig. 5G-K; see also Matsui et al., 1990). In pro- gressively older dermatomal tissue, the location of SlF mRNA expression is shifted from the mediodorsal portion to a more dorsal location in the epithelial dermatome (Fig.

5G,I).

Embryos obtained half a gestational day later, at e11,

revealed persistent SlF mRNA in the telencephalon, but expression in the cranial mesenchyme could no longer be detected. In the trunk, SlF mRNA expression is diminished in cervical somites, whereas the dorsal dermatome of all the posterior somites down to the base of the tail were strongly positive (Fig. 5C). No cells expressing SlF mRNA could be detected in the MSA (Fig. 5D).

At e11.5 SlF mRNA could no longer be detected in the head mesenchyme and the trunk somites. However the expression in the telencephalon and tail somites persisted (Fig. 5E). In addition, SlF mRNA expression was present in the posterior region of the hindlimb bud (Fig. 5F, arrow).

By e12.5, SlF mRNA expression in the telencephalon still persists, and new expression is detected in the interdigital mes- enchyme of the limb buds. Expression in dorsal dermatome was detected only in the tail somites. At this stage, no hybrid- ization signal for SlF mRNA could be detected in dermal mes- enchyme associated with developing somites at more rostral axial levels (not shown).

Fig. 2. Distribution of melanocyte precursors in e11.5 mouse embryos. TRP-2 and c-kit antisense whole-mount in situ hybridization of e11.5 embryos. (A,B) Lateral view of a TRP-2 antisense hybridization. (C) Lateral view of the mid trunk of an embryo hybridized with c-kit antisense probe. (D) Dorsal view of the tail and hindlimb buds of an embryo hybridized with TRP-2 antisense probe. The inset displays cells in the MSA that are at a higher focal level than the cells dispersing laterally on the dorsal migration pathway. (E) Lateral view of the hindlimb region of the embryo shown in A and B. The dashed line roughly indicates the lateral limit of the MSA. Note that many c-kit-positive cells can be detected in the lateral mesenchyme (arrow in C), which can not be detected with the TRP-2 probe (arrow in A). A-B, C-D and E-F

approximately indicate the axial levels of transverse sections shown in Fig. 3. fl, forelimb; hl, hindlimb. Bar, 400 µm in A,B and E; 250 µm in C and 275 µm in D.

17

(19)

736 B. Wehrle-Haller and J. A. Weston

Fig. 3. Melanocyte precursor migration on the lateral pathway. Immunohistochemistry for laminin and fibronectin on transverse sections from an e11.5 TRP-2 antisense whole-mount in situ hybridization. Axial levels of each section are indicated in Fig. 2D. (A,C,E) Bright-field illuminations of sections corresponding to levels A-B, C-D and E-F, respectively. The sections are stained with either anti-laminin antiserum (B,D) or fibronectin antiserum (F). Note: TRP-2-positive cell elongated along a laminin-positive basement membrane (arrow in C and D).

Arrowheads in E point to the penetration limit for the detection of mRNA in that particular embryo. Ln, laminin; Fn, fibronectin; nt, neural tube; dm, dermatome; s, sclerotome; MSA, migration staging area; sg, sensory ganglia. Bar, 25 µm in A-D and 100 µm in E,F.

18

(20)

737 Roles of SlF in melanocyte dispersal and survival Migration and localization of TRP-2 mRNA-

expressing cells is different in embryos

homozygous for Steel alleles compared to wild-type embryos

The Sl null mutation and the less severe Sl

^d

allele were used to reveal the function of SlF during melanocyte precursor migration on the lateral pathway. As above, the hindlimb level of e11.5 embryos was selected to compare melanocyte precursor migration and localization in normal and mutant embryos.

Fig. 6B shows the hindlimb level and the base of the tail of a Sl/Sl (null) embryo. Individual TRP-2 mRNA-expressing cells were detected in the MSA over a length of four to five segments (Fig. 6B, brackets). No cells were found more laterally on top of the dermatome. With the exception of a few cells localized to the mid-hindbrain

border, no other TRP-2 mRNA- expressing cells were found in e11.5 embryos (not shown).

A Sl

^d

/Sl

^d

e11.5 embryo reveals a strikingly different distribution pattern of TRP-2 mRNA-express- ing cells. In addition to cells localized in the MSA, many cells were found dispersed over the somites (Fig. 6C). However, no TRP-2 mRNA-expressing cells were detected anterior to the level of the hindlimb (Fig. 6C), with the exception of some cells at the mid- hindbrain border (not shown). In contrast, in wild-type embryos, a higher density of TRP-2 mRNA- expressing cells were localized over the dermatome compared to the Sl

^d

/Sl

^d

mutant embryos (Fig.

6A).

As a further test of the role of SlF function in promoting melanocyte precursor dispersal, embryos homozygous for the W mutation (phenotypically null for c-kit activity) were examined.

Such embryos showed a pattern of TRP-2 mRNA-expressing cells in the tail region comparable to the pattern observed in Sl/Sl (null) embryos (Fig. 6D).

DISCUSSION

Melanocyte precursors arise in the migration staging area (MSA) prior to their dispersal on the lateral pathway

At e10.5, cells expressing TRP-2 and c-kit mRNA have been reported lateral to neural tube in the head (Steel et al., 1992). In the

trunk c-kit-positive cells have been reported to be lateral to the neural tube between e10.5 and e11 (Manova and Bachvarova 1991). At later stages, we have observed c-kit and TRP-2 mRNA-positive cells at trunk levels posterior to the hindlimb buds (at the base of the tail). In addition to their presence in the MSA of the trunk, melanocyte precursors are known to be present initially in a few regions in the head, including the regions between brain vesicles and posterior to the otic vesicle.

As has been suggested by Steel et al. (1992), it is very likely that these cells are melanocyte precursors and that they co- express both c-kit mRNA and TRP-2 mRNA. It is not yet known, but will be important to learn, what local environmen- tal cues induce the expression of melanocyte specific genes, and when crest cells in the MSA respond to such cues.

Assuming that our marking studies do, in fact, reveal melanocyte precursors, it is of particular interest to note that

Fig. 4. Distribution of melanocyte precursors in e12.5 embryos. TRP-2 antisense whole-mount in situ hybridization of e12.5 mouse embryos. Alternative views are shown of one embryo in A-C.

From B, the embryo is turned clockwise (A) or counter-clockwise (B) to reveal lateral and frontal views of the embryo. The lateral side of a different embryo is shown in D. Note: Arrow in C points to a population of TRP-2-positive cells entering the posterior aspect of the hindlimb. fl, forelimb;

hl, hindlimb. Bar, 500 µm in A-C; 320 µm in D.

19

(21)

738 these cells appear belatedly as a subpopulation among crest- derived cells in the MSA. They seem to arise in the MSA after most of the neurogenic crest-derived cells have dispersed on the medial pathway (see Weston, 1991), but before dispersal

has begun on the lateral crest migration pathway. In this regard, it is also of interest to note that Erickson and Goins (1995;

personal communication) have also recently shown that older crest-derived cells that have become specified as melanocyte B. Wehrle-Haller and J. A. Weston

Fig. 5. Rostrocaudal sequence of SlF mRNA expression in the dorsal dermatome. SlF (KL-M1) antisense whole-mount in situ hybridization of e10.5, e11 and e11.5 mouse embryos and immunohistochemistry for laminin and fibronectin of e10.5 transverse sections. (A,B) Lateral view of two e10.5 embryos, A is slightly younger than B. Arrowheads point to mesenchymal SlF mRNA expression at the posterior border of the first and second branchial arches. (C) Lateral view of an e11 embryo. (D) Dorsal view of an e11 embryo. (E) Lateral view of an e11.5 embryo, a closeup of the tail region is shown in F. Note the expression of SlF mRNA in the posterior aspect of the hindlimb (arrow).

Transverse sections of an e10.5 SlF whole-mount in situ hybridization from a midtrunk (G,H) and cervical (I,K) level. The dermatome shown in G,H is developmentally younger than that in I and K. Arrows mark the limits of the SlF mRNA expression. ba, first branchial arch; hyoid arch; fl, forelimb; hl, hindlimb; d, dermatome; m, myotome; s, sclerotome; nt, neural tube. Bar, 300 µm in A-C, E and F, 470 µm in D and 25 µm in G-K.

20

(22)

739 Roles of SlF in melanocyte dispersal and survival precursors can precociously invade the lateral migration

pathway when they are grafted heterochronically into avian host embryos.

Transient expression of Steel Factor mRNA by dermatomal epithelial cells precedes dispersal of melanocyte precursors on the lateral pathway Whole-mount in situ hybridization with a probe for SlF mRNA has confirmed earlier reports of its presence in the dorsal dermatome (Matsui et al., 1990). In addition, however, this method has revealed the precise expression pattern relative to the pattern of dispersal of putative melanocyte precursors. In particular, our results demonstrate, first, that SlF mRNA expression is transiently localized to the dorsal epithelial dermatome and, second, that this expression precedes the onset of melanocyte precursor dispersal on the migration path towards the dermatome. Finally, our results indicate that SlF mRNA is down regulated in the epithelial dermatome as it de- epithelializes to produce dermal mesenchyme. It is not known how long SlF protein persists in the dermal mesenchyme, or

when SlF mRNA is re-expressed by these cells (Keshet et al., 1991; Motro et al., 1991; Orr-Urtreger et al., 1990). In utero injection of anti-c-kit antibodies (Nishikawa et al., 1991) has revealed a SlF-dependent period at around e14.5, correspond- ing to the time (e13-14) when melanocyte precursors localized in the dermal mesenchyme are known to enter the epidermis (Mayer, 1973). Since melanocyte precursors transiently depend on SlF for survival (see also Morrison-Graham and Weston, 1993), SlF protein must be retained and made acces- sible for the migrating melanocyte precursors in the newly formed dermis. No data are yet available on the presence and location of SlF protein after the transformation of the dermatome into dermal mesenchyme is completed.

Melanocyte precursors in the MSA initially disperse toward the site where SlF mRNA is transiently produced

In the head, at e11 c-kit/TRP-2 mRNA-positive cells localize to regions where SlF mRNA was detected at e10.5 (Steel et al., 1992). Likewise, melanocyte precursors disperse from the MSA

Fig. 6. Melanocyte precursors initially appear but behave differently in Steel mutant embryos. TRP-2 antisense whole-mount in situ hybridization of e11.5 mouse mutant embryos. The tail and hindlimb bud region is shown of embryos being wild type (A, +/+), homozygous for SlF null allele (B, Sl/Sl), homozygous for Sl dickie (C, Sl^d/Sl^d) or homozygous for c-kit functional null allele (D, W/W). The litter that contained embryo D was developmentally slightly younger than other ‘e11.5’ litters so that cells were seen over a wider area along the A-P axis compared to the developmentally older embryo seen in B. In addition, this embryo was stained longer than the other embryos, which affected the appearance of the stained cells. The dashed lines roughly indicate the lateral limits of the MSA. In C, the location of the limb bud, which was removed for genotyping (see Methods), is indicated with parentheses; dirt particles are marked with asterisk. hl, hindlimb; t, tail. Bar, 200 µm.

21

(23)

740 to sites where SlF mRNA had been transiently expressed about 6-12 hours previously, suggesting an influence of SlF protein on the migration and localization of melanocyte precursors.

Interestingly, melanocyte precursors migrating on the lateral pathway seem to follow existing basement membranes, as do the medially migrating crest cells (Tosney et al., 1994). Two extreme models could explain the observed migration pattern.

First, TRP-2/c-kit mRNA-expressing cells could migrate on both the medial and lateral pathways and survive only at locations expressing SlF. This model would imply a passive role of SlF on initial cell migration stressing only the survival effect of SlF on dependent cell populations. Alternatively, cells that express functional c-kit receptors, as implied by c-kit mRNA expression, could be selectively attracted onto the lateral pathway by SlF that is locally produced by dermatomal epithelium and diffuses from its site of expression. This notion is supported by the reports that c-kit-expressing carcinoma cells as well as c-kit transfected endothelial cells show chemotaxis towards SlF in vitro (Blume-Jensen et al., 1991; Sekido et al., 1993), and is consistent with the belated appearance of cells in the MSA that express c-kit and the corresponding delay in onset of migration on the lateral pathway. Although the initial onset of dispersal of melanocyte precursors on the lateral pathway is consistent with a chemotactic response to a source of SlF in the dermatome, the subsequent dispersal of melanocyte precursors that occurs after e11.5 in the dermal mesenchyme of the trunk, limb buds or ventral body wall, cannot be explained by a graded or localized source of SlF.

Melanocyte precursors fail to disperse or survive in Steel null mutant embryos

If SlF were required for initial dispersal of melanocyte pre- cursors, then these cells would be predicted to remain in the MSA in mutants that do not produce this molecule. This pre- diction has been verified by our observations of the behavior of TRP-2 mRNA-expressing cells in Steel null embryos. Such cells appear in the MSA in a timely way, but never disperse, and ultimately disappear. The fate of these melanocyte pre- cursors is not known. Alternatively, their apparent failure to disperse may be the result of failing to survive in the MSA long enough to do so. We know that melanocyte precursors in vitro require transient trophic support from SlF for about 5 days (Morrison-Graham and Weston, 1993). Since the time interval during which melanocyte precursors are generated in the MSA is not known, however, it is difficult to estimate how long indi- vidual TRP-2/c-kit mRNA-expressing cells can survive in the MSA in the absence of SlF. If this period were brief, however, SlF-dependent cells would probably not survive long enough to leave the MSA. In this regard, mast cells, which are also dependent on c-kit/SlF activity, begin to degenerate in vitro within 5 hours after SlF is removed from their culture medium (Iemura et al., 1994; Caceres-Cortes et al., 1994). Survival of individual melanocyte precursors in the MSA of SlF null mutants might be even shorter because these cells would never have received an SlF stimulus.

The conclusion that crest-derived melanocyte precursors require a timely SlF stimulus in the MSA is supported by our observations that TRP-2 mRNA-expressing melanocyte pre- cursors are initially present in the MSA, but are never found on the lateral pathway of embryos that are homozygous for the W mutation. Thus, cells that lack functional c-kit receptors are

unable to detect the presence of SlF, and therefore either fail to survive in the MSA or are unable to depart on the lateral pathway in response to a directional SlF signal.

Soluble SlF is sufficient to permit initial dispersal of melanocyte precursors from the MSA onto the lateral pathway, but not for eventual survival and/or differentiation

The Steel dickie (Sl

^d

) mutation results in a truncated but bio- logically active form of SlF that lacks a transmembrane domain (Brannan et al., 1991). The lack of a transmembrane anchor results in the secretion of SlF by cells from Sl

^d

homozygotes (Flanagan et al., 1991). Since Sl

^d

homozygotes exhibit an inter- mediate phenotype compared to Sl null mutants, the membrane-bound SlF appears to be required for normal devel- opment of these cells.

Paradoxically, neural crest cell cultures are able to give rise to melanocytes in vitro when cultured in the presence of exogenous soluble SlF (Morrison-Graham and Weston, 1993), or on detergent-extracted (cell-free) extracellular matrix (ECM) deposited by wild-type embryonic skin fibroblast in vitro (Morrison-Graham et al., 1990). ECM deposited by such cells seems to contain enough SlF to satisfy the SlF-dependent melanocyte precursors in vitro. In contrast, detergent-extracted ECM deposited by cultured fibroblasts from Sl

^d

homozygotes fail to support melanogenesis in vitro (Morrison-Graham et al., 1990), suggesting that the truncated SlF is not incorporated or retained in ECM secreted by these cells to permit survival and/or differentiation of melanocyte precursors.

Heterochronic grafting experiments (Erickson and Goins, 1995; personal communication) indicate that melanocyte pre- cursors arise late and invade the dorsolateral migration pathway in response to some localized cue(s) in the avian embryo. It seems likely that the soluble form of SlF may be at least one such cue. Thus, in the present report, it is of particu- lar interest that crest-derived melanocyte precursors do, in fact, leave the MSA on the lateral pathway in Sl

^d

homozygotes, sug- gesting that the truncated (soluble) SlF produced by the dorsal dermatome is sufficient for crest cells to initiate dispersal on the lateral pathway. However, since these dispersing melanocyte precursors fail to survive in the dermal mes- enchyme, the truncated SlF appears not to provide an appro- priate survival stimulus in the dermatomal ECM.

Taken together, the behavior and fate of melanocyte pre- cursors in the null mutant compared to that in normal embryos or the Sl

^d

homozygotes suggests that cell-bound and soluble SlF have distinct functions. Although it is not known how much SlF is normally released by dermatomal cells and is present in interstitial crest migration spaces, it seems likely that soluble SlF might be required to promote dispersal of c-kit mRNA-expressing melanocyte precursors, or to attract them to its local source. Likewise, the behavior and fate of melanocyte precursors in Sl

^d

embryos suggests that an ‘immobilized’ form of SlF is required for survival of the responsive cell type. Three published reports support our inference that soluble and cell- bound SlF play distinct roles in cell dispersal and survival, respectively. First, in the head of Sl

^d

embryos, TRP-2 mRNA- expressing cells appear initially and disperse, but then disappear (Steel et al., 1992). Second, germ cells, which are known to require SlF and c-kit for survival can be found in small numbers at the genital ridges in Sl/Sl

^d